Utility of CDX-2 in Distinguishing Between Primary and
Secondary (Intestinal) Mucinous Ovarian Carcinoma
An Immunohistochemical Comparison of 43 Cases
Maria Rosaria Raspollini, MD,* Gianni Amunni, MD,† Alessandro Villanucci, MD,†
Gianna Baroni, BSc,* Antonio Taddei, MD,‡ and Gian Luigi Taddei, MD*
Abstract:Primary and secondary mucinous tumors can involve the ovaries and have similar histologic appearances. The differential di-agnosis is important for surgical and chemotherapeutic treatment and for the prognosis, but often it is extremely difficult. This article dis-cusses an immunohistochemical panel that includes carcinoembry-onic antigen (CEA), cytokeratin (CK) 7, CK20, CA125, CA19.9, and a new marker, CDX-2, for the distinction between primary ovarian mucinous carcinomas and metastatic (intestinal) ovarian tumors. Forty-three cases representing primary and secondary ovarian tumors were considered and consisted of 14 primary mucinous ovarian car-cinomas (PMOCs) and 29 secondary (intestinal) ovarian tumors (SI-OTs). Fisher exact test was performed to evaluate the reliability of the respective antibodies to discriminate between PMOCs and SIOTs. CDX-2 was diffusely positive in all SIOTs and was expressed focally in 3 cases (21.42%) of PMOCs. CK7 was diffusely positive in 13 cases (44.82%) of SIOTs and in 13 cases (92.85%) of PMOCs. CK20 was diffusely positive in 17 cases (58.62%) of SIOTs and in 6 cases (42.85%) of PMOCs. CEA was diffusely positive in 28 cases (96.55%) of SIOTs and in 12 cases (85.71%) of PMOCs. CA19.9 was positive in all SIOTs and in 12 cases (85.71%) of PMOCs. CA125 was positive in 3 cases (10.34%) of SIOTs and in 4 cases (28.57%) of PMOCs. CK7 and especially CDX-2, a specific and sensitive marker, can aid pathologists in making a differential diagnosis (P = 0.003 and
P < 0.0005, respectively), whereas CEA, CK20, CA125, and CA19.9
markers are not high enough to distinguish between primary and sec-ondary mucinous ovarian tumors.
Key Words: CDX-2, ovarian carcinoma, intestinal carcinoma, mu-cinous tumor, metastases
(Appl Immunohistochem Mol Morphol 2004;12:127–131)
M
ucinous metastases to the ovaries are more frequent than primary mucinous ovarian carcinoma (PMOC).1 Pri-mary mucinous carcinomas appear to account for approxi-mately 3.6% of ovarian epithelial neoplasms, 12% of ovarian mucinous neoplasms, and 11% of ovarian carcinomas.2Ovarian mucinous carcinomas occur in patients in the 4th to 7th decades. As many as 63% of the cases are Interna-tional Federation of Gynecology and Obstetrics (FIGO) stage I, and nearly all stage I cases have unilateral ovarian involve-ment.3Therefore, these women do not generally present signs and symptoms of abdominal carcinomatosis, but the clinical presentation is usually that of a large mass and abdominal dis-tension. The proportion of stage I cases varies widely among studies because of the difficulty in distinguishing intraepithe-lial carcinomas from invasive carcinomas.4
Before making a diagnosis of mucinous carcinoma of the ovary, it is imperative to exclude metastatic mucinous car-cinoma. Mucinous adenocarcinomas of the pancreas, biliary tract, colon, appendix, and cervix can mimic ovarian mucinous carcinomas.5–8 Approximately 40% of adenocarcinoma me-tastases to the ovary originate from the colon.9It may be dif-ficult to differentiate between primary and metastatic lesions according to clinical and histopathologic aspects, but the dif-ferential diagnosis is decisive for the surgical and chemothera-peutic treatment and for the prognosis.
Sometimes the clinical history of a patient can help in the exact diagnosis, but a negative anamnesis for neoplastic dis-ease is not helpful, because sometimes metastases can antici-pate a primary tumor. A recent report indicates a rule that cor-rectly classified 90% of the neoplasm. It classifies all bilateral mucinous carcinomas as metastatic and unilateral mucinous carcinomas 10 cm or greater as primary.10
Cdx genes are homeobox genes necessary for intestinal organogenesis and encode for nuclear transcription factors in-volved in proliferation and differentiation of intestinal epithe-lial cells in fetal and adult tissue.11CDX-2 is expressed in nor-mal colonic epithelia and most colorectal adenocarcino-mas.12,13
Received for publication July 29, 2003; accepted September 23, 2003. From the *Department of Human Pathology and Oncology, the †Department
of Gynecology, Perinatology, and Reproductive Medicine, and the §De-partment of Surgery, University of Florence, Florence, Italy.
Reprints: Maria Rosaria Raspollini, MD, Department of Human Pathology and Oncology, University of Florence, viale GB Morgagni, 85. 50134 Florence, Italy (e-mail: [email protected]).
The aim of the present study was to evaluate whether the immunohistochemical expression of CDX-2 in a series of mu-cinous ovarian carcinomas and a series of secondary ovarian carcinoma could be a reliable marker for identifying colorectal metastases from primary ovarian tumors and whether this marker could be added to an immunohistochemical panel of specific markers useful in this differential diagnosis.
MATERIALS AND METHODS
Case Selection
The files of the Department of Human Pathology and Oncology of the University of Florence were searched from 1990 to 2002 for the diagnosis of ovarian mucinous tumors.
From these neoplasias, 14 cases of primary mucinous carcinoma of the ovary and 29 cases of ovarian metastases from intestinal carcinoma were selected to be evaluated in the current analysis. Of these 43 cases, histologic material, clinical data, and follow-up were available. Clinical data available for each patient included age, previous medical history, surgical treatment, disease stage, adjuvant therapy if performed, and follow-up.
All microscopic slides were reviewed without knowl-edge of the previous medical history or the clinical outcome.
The age range of the patients with primary ovarian tumor was between 33 and 84 years (average age, 52.23 years; me-dian, 53 years). None of these women had a medical history of surgical operation for neoplastic disease. The 14 patients un-derwent surgical treatment of abdominal hysterectomy, bilat-eral salpingo-oophorectomy, appendectomy, and omentec-tomy with careful examination of all serosal surface and biop-sies of any suspected lesions. All patients were staged according to the modified staging system of the International Federation of Gynecology and Obstetrics for malignant sur-face epithelial-stromal tumors. Twelve cases were FIGO stage I, and 2 cases were stage III.14Twelve of the 14 cases of pri-mary ovarian carcinomas were unilateral with a size range of 6 to 15 cm, whereas 2 cases had bilateral masses with peritoneal metastasis outside the pelvis. On these 14 patients, a postop-erative chemotherapeutic treatment was performed indepen-dently of the presence or absence of residual disease. The fol-low-up period was 26 to 143 months.
The age range of the patients with ovarian metastases was between 33 and 86 years (average age, 61.31 years; me-dian, 67 years). In 3 of these patients (13.63%), secondary ovarian neoplasias were metachronous in regard to colon car-cinoma. They had undergone surgical treatment of colectomy with regional lymphadenectomy and postoperative chemo-therapeutic treatment of a primary intestinal carcinoma 18, 22, and 30 months previously. Soon afterward, these patients un-derwent surgical treatment of abdominal hysterectomy, bilat-eral salpingo-oophorectomy, and omentectomy for an ovarian mass. In 25 patients (81.81%) with secondary ovarian tumor,
the ovarian metastases were synchronous with intestinal car-cinoma. These 25 women had undergone surgical treatment of colectomy with regional lymphadenectomy, appendectomy, abdominal hysterectomy, bilateral salpingo-oophorectomy, and omentectomy, and subsequent chemotherapeutic treat-ment. In the remaining patient of our series of cases, the sec-ondary ovarian tumor was anticipated in regard to the colon carcinoma. This woman had undergone abdominal hysterec-tomy, bilateral salpingo-oophorechysterec-tomy, appendechysterec-tomy, and omentectomy for an ovarian mass, and 23 months thereafter, she underwent a colectomy with regional lymphadenectomy for intestinal carcinoma.
Immunohistochemistry
The specimens were obtained by surgical resection in all cases and fixed in 10% formalin before being processed in par-affin. For immunohistochemical analysis, a representative sec-tion of each lesion was selected.
The immunohistochemical study was performed using the streptavidin-biotin-peroxidase method (UltraVision kit; Lab Vision, Fremont, CA) with diaminobenzidine as chromo-gen and hematoxylin as nuclear counterstain with antibodies anti–CDX-2 (clone 7C7/D4; Bio-Genex, San Ramon, CA; 1:100 dilution; with Immunostainer Genomix Bio-Genex and MW antigen retrieval for 30⬘ pH 6 with citrate buffer), anticy-tokeratin 20 (clone IT-Ks20.8; Bio-Genex; 1:60 dilution; with Immunostainer Nexes Ventana and protease digestion with phosphate buffer solution), anticytokeratin 7 (clone OV-TL12/30; Bio-Genex; 1:800 dilution; with Immunostainer Nexes Ventana and proteasi antigen retrieval with phosphate buffer solution), anti-CA125 (clone M11; CISbio International France; with Immunostainer Nexes Ventana), anti-CA19.9 (clone NS19.9; CISbio International France; with Immuno-stainer Nexes Ventana), and anticarcinoembryonic antigen (cd66e; clone 12-140-10; Novocastra, UK; 1:500 dilution; with Immunostainer Nexes Ventana and proteasi antigen re-trieval with phosphate buffer solution).
The negative controls were performed by omitting the primary antibodies and by substituting the primary antibodies with nonimmune mouse sera. Appropriate positive and nega-tive controls were run simultaneously. Sections of strongly positive cytokeratin (CK) 7 columnar epithelium cervical, positive CK20 and carcinoembryonic antigen (CEA) colonic adenocarcinoma, positive CDX-2 intestinal epithelium, CA125-positive serous ovarian carcinoma, and CA19.9-positive pancreatic carcinoma were used as CA19.9-positive control tissues for CK7, CK20, CEA, CDX-2, CA125, and CA19.9, respectively.
The immunohistochemically stained sections were evaluated without previous knowledge of the clinical outcome of each patient.
Brown staining of antibody-specific CDX-2 of the nucleus was considered positive. Brown staining of cytoplasm
of antibody-specific CK20 and CK7 was considered positive. Lesions were considered immunoreactive with CEA if the cy-toplasm of columnar epithelial cells showed immunoreactivity equal to glycocalyx staining in intensity. Lesions were consid-ered immunoreactive with CA125 and CA19.9 if they showed staining of the cell membrane, especially on the apical site. The intensity of immunostaining was classified in 3 grades: no staining, weak staining, and strong staining.
Statistical Analysis
Statistical analysis was performed to evaluate the reli-ability of the antibodies to discriminate between primary ovar-ian carcinoma and ovarovar-ian metastases.
The statistical analysis was performed using the Fisher exact test. P valueⱕ0.05 was considered statistically signifi-cant.
RESULTS
CDX-2 was diffusely expressed in the nucleus of all cases of secondary (intestinal) ovarian tumor (100%; Fig. 1), whereas the majority (78.57%) of primary ovarian mucinous carcinomas were CDX-2–negative (Fig. 2), and CDX-2 was expressed focally in the nucleus of only 3 cases (21.42%; Fig. 3) of PMOCs (P < 0.0005). Of these 3 cases of CDX-2 focally positive PMOCs, 2 were positive for CK7, CK20, CEA, and CA19.9 and negative for CA125, and 1 case was negative for CK7, CK20, CEA, CA19.9, and CA125.
The expression of CEA antigen was identical for both types of tumors: it was expressed in both cytoplasm and on the cell membrane strongly in 28 cases of secondary ovarian
tu-mors (96.55%) and 12 cases of PMOCs (85.71%; P = 0.24). CK7 and CK20 showed a cytoplasmic localization in carci-noma cells: CK7 showed positive staining in 13 cases of sec-ondary ovarian tumors (44.82%) and in 13 cases of PMOCs (92.85%), whereas CK20 showed positive staining in 17 cases of secondary ovarian tumors (58.62%) and in 6 cases of PMOCs (42.85%; P = 0.003 and P = 0.51, respectively). CA125 and CA19.9 predominantly showed staining in the cell membrane, especially on the apical site. Occasionally, some expression was found in the cytoplasm. CA125 was expressed
FIGURE 3. CDX-2 was expressed focally in the nucleus of only 3 (21.42%) cases of PMOCs (original magnification⳯40). FIGURE 1. Metastasis of colorectal adenocarcinoma in the
ovary. CDX-2 was diffusely expressed in the nucleus of the secondary (intestinal) ovarian tumor (original magnification ⳯40).
FIGURE 2. The majority (78.57%) of primary ovarian mucin-ous carcinomas were CDX-2–negative (original magnification ⳯40). Sections of secondary (intestinal) ovarian tumors and of primary ovarian carcinoma were run simultaneously.
in 3 cases of secondary ovarian tumors (10.34%) and in 4 cases of PMOCs (28.57%), and CA19.9 was expressed in all cases of secondary ovarian tumors (100%) and in 12 cases of PMOCs (85.71%; P = 0.19 and P = 0.1, respectively).
The results of the immunohistochemical studies are summarized in Table 1.
DISCUSSION
Often the diagnosis of a metastatic tumor to the ovaries is missed by the pathologist because the existence of a present or previous tumor in another side is either not known or disre-garded.15,16Pathologists need to be extremely cautious when evaluating ovarian mucinous tumors, because secondary ovar-ian carcinomas of intestinal origin may closely simulate pri-mary carcinomas of the ovary, and they may be much larger than intestinal primary tumor.17
The overall incidence of metastases of colorectal carci-noma to the ovaries is 10% when malignant ovarian tumors are encountered at the time of surgery.18,19
The histologic distinction between a secondary (intesti-nal) ovarian neoplasm and a primary ovarian mucinous adeno-carcinoma can be extremely difficult and sometimes impos-sible: as many as 45% of secondary ovarian tumors are clini-cally seen to be primary ovarian carcinomas, and many are misinterpreted as such on pathologic examination, even when there is a known intestinal carcinoma.5
The difficulty of differential diagnosis on histologic grounds alone between primary ovarian carcinoma and colo-rectal metastases to ovaries has prompted an intensive search for immunohistochemical markers that can help distinguish between these 2 lesions.
Carcinoembryonic antigen is a highly glycosylated cell surface protein overexpressed in a variety of human tumors, such as colorectal, gastric, pancreatic, ovarian, breast, and nonsmall cell lung carcinomas.20It is characterized by an on-cofetal pattern of developmental expression. Several studies have demonstrated that CEA immunostaining is of no value in the differentiation between secondary ovarian tumors showing a mucinous pattern and PMOC, because both show equally strong staining.5,21
CA125 and CA19.9 are not discriminatory markers be-tween ovarian mucinous carcinoma and intestinal carci-noma.21–23
Cytokeratin 7 and CK20 have been suggested as immu-nohistochemical discriminators between ovarian masses of müllerian type (CK7+, CK20–) and of intestinal type (CK7–, CK20+), but whereas the discrimination of the nonmucinous carcinomas does not raise any problems, the distinction be-tween a mucinous colonic and a mucinous ovarian carcinoma is more difficult. A few mucinous ovarian tumors resemble intestinal carcinomas in their immunohistochemical staining pattern (CEA+, CK20+, CK7–), and some colonic carcinomas expressed CK7.21A recent study reported that there are metas-tases from colorectal carcinomas with CK7+/CK20– immuno-phenotype and with CK7+/CK20+ immunoimmuno-phenotype, so these immunohistochemical stainings are not helpful for the differential diagnosis.24An important element in the evalua-tion of positive or negative staining is that focal expression of CK7 and CK20 should be given the same significance as nega-tive immunostaining because of the steady increase of anti-body sensitivity as a result of improved pretreatment meth-ods.25
From these data, it was discovered that the specificity of CEA, CK20, CK7, CA125, and CA19.9 markers is not suffi-ciently high in distinguishing secondary (intestinal) ovarian tu-mors from primary ovarian mucinous carcinomas. The current study, which combines all 3—morphologic data, clinical data, and immunohistochemical assessment of malignant tumors of the ovary—was undertaken to determine whether the introduc-tion of a new monoclonal antibody to an immunohistochemi-cal panel, such as CDX-2, can help us in the differential diag-nosis of mucinous ovarian carcinoma from the most frequent CDX-2–positive secondary (intestinal) ovarian lesions.26
In a recent study on metastases to the lungs, Barbareschi et al24analyzed 5 cases of secondary pulmonary mucinous
tu-mors from ovarian neoplasia that were all CDX-2–positive, and Werling et al27found CDX-2 high-level expression in
mu-cinous ovarian neoplasias that were also villin-positive. On the other hand, our immunohistochemical data show that CDX-2 is a highly sensitive and specific marker to add in a limited immunohistochemical panel for distinguishing CDX-2–negative primary ovarian mucinous carcinomas from CDX-2–positive secondary (intestinal) ovarian tumors.
These immunohistochemical results may be of impor-tant clinical significance because most women with mucinous ovarian tumors have a hidden intestinal carcinoma or have
al-TABLE 1. Immunohistochemistry Analysis in Ovarian Mucinous Tumors
Tumor CDX-2+ P/T (%) CK7+ P/T (%) CK20+ P/T (%) CEA+ P/T (%) CA125+ P/T (%) CA19.9+ P/T (%) Primary ovarian carcinomas 3/14 (21.42) 13/14 (92.85) 6/14 (42.85) 12/14 (85.71) 4/14 (28.57) 12/14 (85.71) Secondary (intestinal) ovarian tumors 29/29 (100) 13/29 (44.82) 17/29 (58.62) 28/29 (96.55) 3/29 (10.34) 29/29 (100)
ready undergone surgical treatment of a previous colorectal tumor.28
The exact histologic diagnosis permits the best surgical and chemotherapeutic treatment of the individual patient. These results indicate the necessity, nevertheless, to investi-gate the CDX-2 marker on a larger series of mucinous ovarian lesions with known clinical history.
REFERENCES
1. Riopel MA, Ronnett B, Kurman RJ. Evaluation of diagnostic criteria and behavior of ovarian intestinal-type mucinous tumours: atypical prolifera-tive (borderline) tumors and intraepithelial, microinvasive, invasive, and metastatic carcinomas. Am J Surg Pathol. 1999;23:617–35.
2. Kurman RJ, ed. Blaustein’s pathology of the female genital tract. 5th ed. New York: Springer-Verlag, 2001.
3. Hoerl HD, Hart WR. Primary ovarian mucinous cystoadenocarcinomas: a clinicopathologic study of 49 cases with long-term follow-up. Am J Surg Pathol. 1998;22:1449–62.
4. Wartkin W, Silva EG, Gershenson DM. Mucinous carcinoma of the ovary: pathologic prognostic factors. Cancer. 1992;69:208–12. 5. Lash RH, Hart WR. Intestinal adenocarcinomas metastatic to the ovaries:
a clinicopathological evaluation of 22 cases. Am J Surg Pathol. 1987;11: 114–21.
6. Young RH, Hart WR. Metastases from carcinomas of the pancreas simu-lating primary mucinous tumors of the ovary: a report of seven cases. Am J Surg Pathol. 1989;13:748–56.
7. Daya D, Nazerali L, Frank GL. Metastatic ovarian carcinoma of large intestinal origin simulating primary ovarian carcinoma: a clinicopatho-logic study of 25 cases. Am J Surg Pathol. 1992;97:751–8.
8. Ronnett BM, Kurman RJ, Shmookler BM, et al. The morphologic spec-trum of ovarian metastases of appendiceal adenocarcinomas: a clinico-pathologic and immunohistochemical analysis of tumors often misinter-preted as primary ovarian tumors or metastatic tumors from other gastro-intestinal sites. Am J Surg Pathol. 1997;21:1144–55.
9. Wauters CCAP, Smedts F, Gerrits LGM, et al. Keratin 7 and 20 as diag-nostic markers of carcinomas metastatic to the ovary. Hum Pathol. 1995; 26:852–5.
10. Seidman JD, Kurman RJ, Ronnett BM. Primary and metastatic mucinous adenocarcinomas in the ovaries: incidence in routine practice with a new approach to improve intraoperative diagnosis. Am J Surg Pathol. 2003; 27:985–93.
11. Suh E, Traber PG. An intestine-specific homeobox gene regulates prolif-eration and differentiation. Mol Cell Biol. 1996;16:619–25.
12. Hinoi T, Tani M, Lucas PC, et al. Loss of CDX-2 expression and micro-satellite instability are prominent features of large cell minimally differ-entiated carcinomas of the colon. Am J Pathol. 2001;159:2239–48. 13. Qualtrough D, Hinoi T, Fearon E, et al. Expression of CDX-2 in normal
and neoplastic human colon tissue and during differentiation of an in vitro model system. Gut. 2002;51:184–90.
14. American Joint Committee on Cancer AJCG Staging Manual, 5th ed. Philadelphia: Lippincott Raven, 1997.
15. Woodruff JD, Murty YS, Bhaskar TN, et al. Metastatic ovarian tumor. Am J Obstet Gynecol. 1970;107:202–9.
16. Yazigi R, Sandstad J. Ovarian involvement in extragenital cancer. Gyne-col OnGyne-col. 1989;34:84–7.
17. Miller BE, Pittman B, Wan JY, et al. Colon cancer with metastasis to the ovary at time of initial diagnosis. Gynecol Oncol. 1997;66:368–71. 18. Graffner HOL, Alm POA, Oscarson JEA. Prophylactic oophorectomy in
colorectal carcinoma. Am J Surg. 1983;146:233–5.
19. Abu-Rustum NR, Barakat RR, Curtin JP. Ovarian and uterine disease in women with colorectal cancer. Obstet Gynecol. 1997;89:85–7. 20. Sikorska H, Shuter J, Gold P. Clinical applications of carcinoembryonic
antigen. Cancer Detect Prevent. 1988;12:321–55.
21. Sheahan K, O’Brien MJ, Burke B. Differential reactivities of carcinoem-bryonic antigen (CEA) and CEA-related monoclonal and polyclonal an-tibodies in common epithelial malignancies. Am J Surg Pathol. 1990;94: 157–64.
22. Loy TS, Quesenberry JT, Sharp SC, et al. Distribution of CA125 in ad-enocarcinomas: an immunohistochemical study of 481 cases. Am J Clin Pathol. 1992;98:175–9.
23. Moll R, Pitz S, Levy R, et al. Complexity of expression of intermediate filament proteins, including glial filament, in endometrial and ovarian car-cinomas. Hum Pathol. 1991;22:989–1001.
24. Barbareschi M, Murer B, Colby TV, et al. CDX-2 homeobox gene expres-sion is a reliable marker of colorectal adenocarcinoma metastases to the lung. Am J Surg Pathol. 2003;27:141–9.
25. Yaziji H, Davie PL, Gown AM, et al. On the use of cytokeratins 7 and 20 in the differentiation between transitional cell carcinoma and prostatic adenocarcinoma [letter]. Am J Clin Pathol. 2001;115:159–60.
26. Beck F, Chawengsaksophak K, Waring P, et al. Reprogramming of intes-tinal differentiation and intercalary regeneration in Cdx2 mutant mice. Proc Natl Acad Sci U S A. 1999;96:7318–23.
27. Werling RW, Yaziji H, Bacchi CE, et al. CDX2, a highly sensitive and specific marker of adenocarcinomas of intestinal origin: an immunohis-tochemical survey of 476 primary and metastatic carcinomas. Am J Surg Pathol. 2003;27:303–10.
28. Rodriguez IM, Prat J. Mucinous tumour of the ovary: a clinicopathologic analysis of 75 borderline tumors (of intestinal type) and carcinoma. Am J Surg Pathol. 2002;26:139–52.
